Budget Optimization

# Budget Optimization

Part of [Agent Skills™](https://github.com/itallstartedwithaidea/agent-skills) by [googleadsagent.ai™](https://googleadsagent.ai)

## Description

The Budget Optimization skill applies AI-driven forecasting and portfolio theory to allocate advertising budgets across campaigns for maximum return. Rather than treating each campaign as an isolated budget silo, this skill models the entire account as an investment portfolio, dynamically shifting spend toward campaigns with the highest marginal return on ad spend (ROAS) while respecting business constraints like minimum brand presence and geographic coverage.

The optimization engine ingests historical performance data, seasonality patterns, competitive dynamics, and conversion lag curves to build predictive models for each campaign's response to budget changes. It identifies campaigns that are impression-share-limited (underfunded relative to demand), campaigns with diminishing returns (overfunded past the efficient frontier), and campaigns where budget shifts would yield measurable incremental conversions.

Portfolio bidding strategies are a key lever. The skill evaluates whether campaigns should use individual or shared budgets, whether portfolio bid strategies can aggregate conversion signals across thin-data campaigns, and how seasonal budget adjustments should be timed relative to demand curves. It produces actionable budget reallocation plans with expected impact projections and confidence intervals.

## Use When

- User asks to "optimize my budget" or "allocate budget better"
- User mentions "budget pacing" or "campaigns running out of budget"
- User wants to know "where to increase spend" or "where to cut budget"
- User asks about "shared budgets" or "portfolio bidding"
- User mentions "ROAS optimization" or "maximize conversions within budget"
- User asks about "seasonal budget adjustments" or "budget planning"
- User wants to "reduce wasted spend" or "improve budget efficiency"
- User mentions "impression share lost to budget"

## Architecture

```mermaid
flowchart TD
    A[Historical Performance Data] --> B[Data Preparation]
    C[Seasonality Signals] --> B
    D[Competitive Landscape] --> B

    B --> E[Campaign Response Modeling]
    E --> F[Marginal CPA Curves]
    E --> G[Impression Share Opportunity]
    E --> H[Conversion Lag Analysis]

    F --> I[Portfolio Optimizer]
    G --> I
    H --> I

    J[Business Constraints] --> I
    K[Total Budget Envelope] --> I

    I --> L{Optimization Strategy}
    L --> M[Shift to High ROAS]
    L --> N[Fund Impression Share Gaps]
    L --> O[Seasonal Pre-Allocation]
    L --> P[Portfolio Bid Strategy Setup]

    M --> Q[Budget Reallocation Plan]
    N --> Q
    O --> Q
    P --> Q

    Q --> R[Expected Impact Projection]
    Q --> S[Implementation Schedule]
    Q --> T[Monitoring Thresholds]
```

## Implementation

Budget optimization engine with marginal return modeling:

```javascript
async function optimizeBudgets(customerId, config) {
  const { totalBudget, lookbackDays = 90, constraints = {} } = config;

  const campaignData = await getCampaignPerformance(customerId, lookbackDays);
  const seasonalFactors = calculateSeasonalFactors(campaignData);
  const marginalCurves = buildMarginalReturnCurves(campaignData);

  const currentAllocation = campaignData.map(c => ({
    campaignId: c.id,
    name: c.name,
    currentBudget: c.dailyBudget,
    spend: c.avgDailySpend,
    conversions: c.avgDailyConversions,
    roas: c.conversionValue / c.cost,
    impressionShareLostBudget: c.isLostBudget,
    marginalCPA: marginalCurves[c.id].marginalCPA
  }));

  return portfolioOptimize(currentAllocation, totalBudget, constraints);
}

function buildMarginalReturnCurves(campaignData) {
  const curves = {};

  for (con